Return to Level1Techs.com

The Subject of SSD Reliability in 2019: Advice Wanted/ Plus my Observations as a Tech

helpdesk
ssd
recommendations
#1

I searched for a thread containing the information I’m looking for, but could not find a current one. If you know of one please let me know and I will add the link in this post.

In this thread, I want to discuss:

  • Reccomended brands/models of SSDs to use in 2019.

  • Observations and statistics that you have found in your deployments.

My company has predominantly installed Adata SSDs in all of our client’s computers for the last 8 years, both business and residential. 80% of our trusted Adata drives that were deployed 8 years ago are still fully functional. However, over the last 9 months, we have experienced a obscene amount of SSD failures (almost 100% failure rates). The new drives we order have been failing within 2 months, if not out of the box. We quickly switched brands and started using Kingston A400 SSDs. These work, but we have already had several die. There has been in the past issues with Kingston’s firmware on these drives, but we verify that they are running the lastest version.

I am aware that there are business/server class SSD available. However, the cost is not justifiable for either residential machines, or business machines that rely on servers for data storage and processing.

We have only used a limited number of other brands in the past, but the failure rates have proven to be far too high. (I will look up said brands and insert the models and dates that they were used along with approximate failure rates)

I am looking for recommendations on what brand/model to purchase that is cost effective. I am especially interested to hear from those of you who have statistics on large numbers of SSD deployed (say like 50 of the same brand over time).

I noticed that Western Digital has stepped into the SSD market, I purchased a 500GB drive from them for my personal use about 6 months ago, but I do not have enough statistics on their reliability.

–Ethernet_Warrior

Lawrence Systems Forums version of this thread:

1 Like

#2

Have had Samsung drives since 830 ( almost one of every type too) have had zero issues

0 Likes

#3

I’ve heard good things about the Samsung drives. I’ve owned a couple. However they are double the price. ($58 for a 240GB. I’d like to find an option for less than $40, we pay $27 for the Adatas rn)

0 Likes

#4

Wd blue drives aren’t bad

0 Likes

#5

Well I believe the old saying still hold’s true. You get what you pay for. I assume these clients are just typical desktop use?
Are they maxing these drives out and or going over 80% constantly?
I think crucial and Toshiba may have some decent drives.
Have not followed much on pricing.

0 Likes

#6

Agreed. There’s plenty of $15 SSDs from no-name brands that should be avoided.

Yes, most of our client run office type software, so we don’t need anything high performance. Just an acceptable amount of reliability.

0 Likes

#7

What model of adata have ya been buying?

0 Likes

#8

Terms like “obscene” and “100%” failure rates fly in the face of statistical probability. If they truly are that extreme, then I would look to a configuration option which is constantly re-writing a file (likely a log file) and essentially drilling a hole in your flash.

2 Likes

#9

SU650 and SU800, usually 240GBs or 500GBs

0 Likes

#10

I’m not sure I follow. Are you saying that might be the cause of the issue?

0 Likes

#11

When we set up disk images in schools for deployment to hundreds of desktops, we had security and monitoring software installed. Some of those logged user activity every few seconds — to the same file(s). Due to the way flash works (specifically write amplification) the damage ended up being quite severe. Since the same software was deployed to all desktops, and all desktops were on for the same amount of time each day, the logging itself cause a cluster of failures. You may be experiencing a similar effect.

1 Like

#12

Ah, that makes sense. I will look into this some more. I’m not sure it would explain all of the failures. We deploy to a variety of of clients.

0 Likes

#13

It’s something to think about at least.

Another thing that caused us issues was a software driver incompatibility. I think it was for a network printer. All of the clients tried to connect to the printer, failed, generated error logs, then tried again less than a minute later, continuously, all day, every day, for months before anyone noticed.

Unusual levels of disk writes can be generated in all sorts of places.

0 Likes

#14

I shall look into that. I hadn’t thought to track where and in what situations we have the most drives fail.

That said, we are only having new drives fail. The old ones last forever. :thinking:

0 Likes

#15

Do ya order these in bulk? Same retail store?

0 Likes

#16

Tight clusters of drive failures — in my experience — tend to track back to common pieces of software that hammer storage. What, why and how that happens varies on a case-by-case basis, but user activity is rarely responsible. If you can audit and/or monitor I/O activity on your storage (specifically writes for SSDs) then you can track down the common factor relatively quickly.

At school we had a base image for each hardware profile, and that was customised for each faculty. Some software was common across different images, but once we knew we had a problem it generally didn’t take long to work out what software on which image(s) was causing the problem.

Generate S.M.A.R.T. reports on your drives once a day, collate them on a server, track relevant fields over time, graph them, and look out for trend lines that go ballistic. That’s what S.M.A.R.T. is there for, after all — may as well use it.

1 Like

#17

How would you figure out what is really doing the hammering in windows/linux? Are there utilities that can be used to track the magnitude of writes each process is doing?

0 Likes

#18

For what your business does, I would only buy samsung evo’s or toshiba’s.

1 Like

#19

I have multiple samsung 850, 860 and 970 EVO (purchased over last 2-3 years) and have yet to have a failure.

I’m confident enough in them that my primary PC is ALL SSD now, with ~2.5 TB split between 3 samsung drives in it.

I also have a 240 GB OCZ from say 2015 that is still doing fine.

Anecdotal, but there you go.

I agree you should probably buy samsung. SSD pricing is very tight (you’re not going to save heaps per unit by whatever you buy), all SATA ones basically saturate the bus, and samsung has the best warranty. I’d suggest that indicates they are confident enough in their product to not have to field excessive claims.

0 Likes

#20

If not Samsung, theres crucial, mushkin, pny, SanDisk. I have some old Kingston v300 and I really wouldn’t recommend giving Kingston money ever again. There should have been a class action lawsuit over that.

Id still pick Samsung over those but I’d put crucial right up there with them. I have bought all those brands without issues though.

1 Like